The results of Chapter 6 are discussed in this chapter. The headlines correspond to the headlines in Chapter 6.
7.1 Medium voltage variations without feeders
Results from this test condition would indicate that the OLTC fixed set point control works well in case of MV variations. If set point is set at 230 V, it allows the full tolerated ± 10%
voltage range to be used within a LV network. In a traditional fixed tap ratio transformer, part of that allowed ± 10% voltage range is also used in a MV network. A fixed set point control method would work well, for example at the furthest secondary substation in ra-dial MV network, where MV networks voltage variations would be the greatest.
7.2 One feeder with load or production
Traditional network planning is based on two extreme cases. Maximum load and mini-mum production and vice versa. Chapter 6.2 test conditions were made to compare two different control methods in these situations. Although limits of maximum cases were restricted by the equipment available in the laboratory, results enable the comparison between the two control methods.
Whether the control parameters of the fixed set point control could be chosen in a way that the control method would be able to detect voltage violation that occurs far away from the supply transformer, depends entirely of the situation. The parameters of the fixed set point control should be set in such a way that voltage variation caused by ex-treme load or production that occurs far away from the supply transformer is detected as a voltage variation at the supply transformer. However, an OLTC is discrete component and therefore the minimum possible tolerated voltage bandwidth at the secondary side of the transformer is limited by voltage difference between two steps. Control parameters of fixed set point control would be fixed in one fits all principle. Therefore, situation where overall loading condition of network would be low but loading condition at far away from the supply transformer would be high, the effect to the voltage at the supply transformer might not be high enough.
7.3 Two different feeders in load and in production
Results in Chapter 6.2 demonstrate behaviour in case of one feeder, but real life LV networks have multiple feeders. Within those feeders, one has lowest voltage and one with highest. In test conditions in Chapter 6.3 these two extreme feeders are considered.
Voltages of other the feeders would be something in between these two. However, other feeders would contribute to the voltage drop or rise at the secondary side of the trans-former. If the load of other feeders would accumulate with load of the feeder with mini-mum voltage, this would affect the behaviour of fixed set point control, because con-trolled voltage would have higher variation. This would affect behaviour in both Chapters 6.2 and 6.3.
7.4 Two different feeders in high load and in high production, so that the voltage difference exceeds limits of the control method
Results in Chapter 6.3 demonstrated that in a condition with two different feeders in load and production the CVC is able to solve voltage violation that would be left unnoticed by the fixed set point control. However, if voltage difference between the maximum and the minimum voltage at the network is too high, the CVC cannot solve the situation. Test condition in chapter 6.4 was made to demonstrate this situation.
The BES voltage control attempts to bring voltage at the connection point of the BES closer to the nominal voltage parameter of the voltage control of BES. This voltage con-trol of BES works, if the nominal voltage of network is set correctly. Assumption that 230 V is the best possible solution, might not fit all networks. In a network, that has high load and relatively not so high production, higher nominal voltage parameter for voltage con-trol of the BES would improve the voltage concon-trol.
The voltage control of the BES is able to contribute correctly to voltage control, if the BES is in the same feeder with voltage rise problem and the voltage at the connection point of the BES is above the nominal voltage or if the BES is in the same feeder with voltage drop problem and the voltage at the connection point of the BES is below the nominal voltage. This would be the general case. However, one possible exception to this would be a network that has a feeder with high load close to the secondary substa-tion and high producsubsta-tion at the end of the feeder. If the BES would be connected close or to the same connection point as high load, the voltage at the connection point of the load could be below the nominal voltage at the same time the rest of the same feeder
has voltage rise problem. In this case, the BES voltage control would bring voltage at the connection point closer to nominal voltage, which would exaggerate the voltage rise ef-fect.
7.5 Medium voltage variations with two different feeders in load and in production
Where experiments in Chapter 6.1 enabled comparison between two control methods in similar actions, experiments in Chapter 6.5 enable better comparison whether action of either control methods are better than others.
Results in Figure 52, Figure 53, Figure 56 and Figure 57 demonstrate effect of MV volt-age rise, from which latter two MV voltvolt-age rise is higher. In these experiments, actions of both control methods is similar. However, this is not the cases with voltage drop.
Results in Figure 54, Figure 55, Figure 58 and Figure 59 demonstrate effect of MV volt-age drop, from which latter two MV voltvolt-age drop is higher. Actions of two control algo-rithms differs here.
Fact that this happened on the MV voltage drop and not rise, depends on the situation at network. Anyway, from this we can see that in some cases the CVC can save unnec-essary steps of the OLTC in MV voltage variation situations.
7.6 Critique for the CVC and active voltage control
The approach of this thesis with direct measurement in strategic locations has some drawbacks. Creating new measurement infrastructure that covers whole network solely for voltage control purpose can be costly. Creating new measurement infrastructure in strategic locations is vulnerable to network topology change. [7] However, this problem would be solved, if real time smart metering data could be utilized. In comparison to measurement devices used in thesis, smart metering might not be as accurate, and therefore increase margin. Other option is to use state estimation [7].
This CVC method does not have cooperation with the HV/MV OLTC, which can result in unnecessary tap operations and voltage fluctuations at a customer supply point. Coordi-nation can be done by having different timers for cascading OLTCs [8]. Having higher tolerance at HV/MV side than at MV/LV side exposes MV/LV transformer to unnecessary tap operations. Having higher tolerance at MV/LV side than at HV/MV side minimizes unnecessary tap operations, but disturbances at a LV network last longer [7]. Which of last two cases would be more efficient depends on whether tolerated voltage limits stated
in European standard EN-50160 are reached. If tolerated voltage limits are not achieved, the tolerance of MV/LV control need to be lowered in an expense of unnecessary tap operations. Another approach is to create system with communication between two OLTCs. This enables more efficient control methods, which can for example reduce volt-age restoration time of customers. [8]
In passive voltage control method, a distribution system operator owns all the network components. If distribution system operator would change to active voltage management that would utilize BES and DG, this would not be the case. This would create uncertainty that whether system operator can trust that these resources are available any given time [1]. Also, in real life situation main operation purpose of a BES is unlikely to be voltage control. Voltage control could serve as an additional feature. A BES would unlikely be owned by the network operator as well, so compensation of using a BES for voltage control should be considered. Scalability of the BES for voltage control is also a chal-lenge. The amount of energy storage capacity needed to have ever higher impact to the voltages amplitude has diminishing return effect [20] .
Despite the DSO does not own the DG or the BES, there is three possible ways to con-nect the DG or BES and the DSO. The first possibility is that the voltage control function-ality can be in grid codes. The second possibility is an agreement that when installing new DG, the owner of the DG and the DSO agree to use the DG for voltage control. This increases hosting capacity of DG. The alternative would be to not have this agreement, and voltage rise effect would results in curtailment of production. The third possibility is a third party that can offer aggregated voltage control resources as a voltage control service for the DSO. [7]
Implementing active voltage control for first time would be laborious for distribution sys-tem operator as the network planning tools are not currently able to consider different voltage control strategies. [1]
Voltage rise effect is a limiting factor for a LV network DG hosting capacity. However, it is anticipated that voltage imbalances will become limiting factor. An OLTC is unable to reduce voltage imbalances. [21] However, in Finland the customers have mainly three phase loads.
The regulatory environment does not encourage active voltage control. In Finland distri-bution system operators are obligated to connect DG to network, but not in the most cost-efficient way. The regulatory environment allows capital expenditures but no in-crease of operational expenditures. The active voltage control would save capital ex-penditures but in expense of increasing operational expenses. [1]
7.7 Future work
Further development of the CVC includes:
Implementation of a DG as a part of the CVC.
Reactive power control of the BES for primary way to affect voltage.
Taking into account operation of an OLTC at a primary substation.
More precise definition of margin used in the CVC.
Adding memory to the control of the CVC.
The CVC in this thesis included voltage control of a BES, but the reactive power control potential was not used. This could be utilized. In addition to BES, a DG could be imple-mented as a part of the CVC. Another component taken into account could be the OLTC at the primary substation, which could make the cooperation of the OLTCs more efficient.
Future development could also include more precise definition of margin used in the CVC. This could be done, for example based on voltage sensitivity analysis of the net-work. Another improvement to the algorithm would be adding memory to the deadband filter. In case of communication error or fault in grid, the CVC algorithm has a deadband filter, which neglects the faulty measurement. However, possible missing measurement of most significant measurement point, exposes algorithm to faulty actions. The CVC could be improved adding memory to the control.
Further validation of the CVC would require simulations in variety of grid conditions. After this case study in actual network would be needed.